数值模式预报性能的地域性特点初步分析

Analysis on Geography & Physiognomy Character of Numerical Model Prediction Performance

根据地理地貌,将全国分成东北、西北、东南、西南四个区域,对T639、T213、ECMWF、日本和德国数值预报模式2008年5—9月的预报分区域进行了客观检验分析。结果表明:模式的降水预报能力与地理的关系似乎比与地形的关系更为紧密一些;2008年,各模式在西南地区的降水预报能力随着降水级别的增加而减弱,到暴雨级别,西南区已成为各模式Ts评分最低区域;降水级别预报正确率高值区域在北方两区,正确率最低值出现在四川盆地;模式降水量预报平均误差显示,大误差并不出现在高原主体,而基本都与大的山脉对应,且都为负误差,但四川盆地除外,仍然是大误差所在地;2 m温度预报平均误差的量级是随海拔高度的增加而增加,误差由正转为负;各模式高空要素预报最大误差大都出现在西南区,其次是西北区。

China area is divided into four regions,namely northeast,northwest,southeast and southwest according to geography and physiognomy to analyze the relationship between the geography physiognomy and the predicting capability of models T639,T213,Japan,Germany and ECMWF during May to September in 2008.The results show that model predicting capability of rainfall is more related to geography than to physiognomy.In the southwestern China all the models＇ predicting capability of rainfall declines as the increase of precipitation,and the models gain the lowest Ts grade estimation for the rainstorm.The higher accurate rate of rain grade prediction appears in some regions of the northern China,and the lowest accurate rate appears in Sichuan Basin.The mean error of the model precipitation prediction shows that instead of occurring on the plateaus,the big errors of rainfall prediction almost appear in the same places as the large mountains and the errors are negative ones with an exception in Sichuan Basin where big errors still exist.Unlike rain prediction,the prediction error of temperature at 2 meter seems to be closely connected with the altitude above sea level.The average error shifts from the positive to the negative with the absolute value aggrandizement as height increase;the prediction errors of upper air elements reach their peaks in the southwestern China,and then in the northwestern China.